Method and device for determining process parameters of printed circuit board assembly, equipment and medium

Abstract

The invention discloses a process parameter determination method and device of a printed circuit board assembly, equipment and a medium, and relates to the technical field of workshop production, and the method comprises the steps: obtaining defect data of a target printed circuit board assembly; respectively inputting the defect data into a pre-trained patch parameter prediction model and a pre-trained printing parameter prediction model to obtain initial patch operation parameters and initial printing operation parameters; through a preset parameter optimization model obtained through training based on a multi-task learning framework model, optimizing the initial surface mounting operation parameter and the initial printing operation parameter according to the internal relation between the surface mounting operation parameter and the printing operation parameter, and obtaining a target surface mounting operation parameter and a target printing operation parameter; and adjusting the surface mounting operation parameter and the printing operation parameter of the printed circuit board assembly production system to be the target surface mounting operation parameter and the target printing operation parameter. According to the invention, the production optimization efficiency and the production optimization effect of the printed circuit board assembly product can be improved.

Description

Technical Field

[0001] The present application relates to the field of workshop production control technology, and in particular to a method, device, equipment and medium for determining process parameters of a printed circuit board assembly. Background Art

[0002] As one of the core links in the assembly of electronic products, the SMT (Surface Mount Technology) workshop is usually equipped with printing machines, placement machines, reflow soldering, AOI (Automated Optical Inspection) and other equipment. These equipment are mainly responsible for mounting electronic components (such as chips, resistors, capacitors, etc.) on printed circuit boards (PCB) to form printed circuit board assemblies (PCBA).

[0003] However, each device in the SMT workshop operates independently, and the parameters of each device fluctuate greatly during the production of printed circuit board assembly products. On this basis, in order to ensure the consistency and quality of the produced printed circuit board assembly products, manual debugging is usually performed based on the operator's experience, which is not only inefficient for the production optimization of printed circuit board assembly products, but also difficult to achieve a good production optimization effect. Summary of the invention

[0004] The main purpose of the present application is to provide a method, device, equipment and medium for determining process parameters of a printed circuit board assembly, aiming to improve the production optimization efficiency and production optimization effect of printed circuit board assembly products.

[0005] To achieve the above object, the present application provides a method for determining process parameters of a printed circuit board assembly, the method comprising:

[0006] acquiring defect data of a target printed circuit board assembly;

[0007] Inputting the defect data into a pre-trained patch parameter prediction model and a printing parameter prediction model respectively to obtain initial patch operation parameters and initial printing operation parameters;

[0008] By using a preset parameter optimization model trained based on a multi-task learning framework model, the initial patch operation parameters and the initial printing operation parameters are optimized according to the intrinsic relationship between the patch operation parameters and the printing operation parameters to obtain target patch operation parameters and target printing operation parameters;

[0009] The mounting operation parameters and the printing operation parameters of the printed circuit board assembly production system are adjusted to the target mounting operation parameters and the target printing operation parameters.

[0010] In one embodiment, the step of obtaining defect data of a target printed circuit board assembly comprises:

[0011] acquiring an image of the target printed circuit board assembly;

[0012] Inputting the image of the target printed circuit board assembly into a pre-trained defect detection model to obtain defect data of the target printed circuit board assembly;

[0013] Among them, the defect detection model is trained based on the model to be trained, the backbone network of the model to be trained introduces a variable kernel convolution, and the positioning loss function of the model to be trained is a pixel intersection-over-union loss function.

[0014] In one embodiment, the training process of the defect detection model includes:

[0015] Obtaining a sample set of defect data of a printed circuit board assembly;

[0016] Freezing a first preset number of convolutional layers close to an input layer in the model to be trained;

[0017] According to the defect data sample set, performing iterative training for a second preset number of training cycles on the frozen model to be trained to obtain an initial detection model;

[0018] The initial detection model is unfrozen, and based on the defect data sample set, the unfrozen initial detection model is iteratively trained for a third preset number of training cycles to obtain the defect detection model.

[0019] In one embodiment, the step of obtaining a sample set of defect data of a printed circuit board assembly further includes:

[0020] Performing data enhancement processing on the defect data sample set to obtain a defect data sample set after data enhancement processing.

[0021] In one embodiment, the step of obtaining a defect data sample set of a printed circuit board assembly includes:

[0022] acquiring a plurality of defect image samples of a printed circuit board assembly;

[0023] Annotating the defect area and defect type of each defect image sample to construct an initial defect data set of the printed circuit board assembly;

[0024] Performing size unification processing on each image included in the initial defect data set;

[0025] The initial defect data set after size normalization is subjected to data enhancement processing to obtain the defect data sample set.

[0026] In one embodiment, before the step of inputting the defect data into a pre-trained patch parameter prediction model and a printing parameter prediction model to obtain initial patch operation parameters and initial printing operation parameters, the method further includes:

[0027] Obtaining a sample set of defect data of a printed circuit board assembly, a sample set of mounting operation parameters of a printed circuit board assembly production system, and a sample set of printing operation parameters;

[0028] Using the defect data sample set as input of a preset decision tree model, and using the patch operation parameter sample set as output of the decision tree model, so as to construct the patch parameter prediction model;

[0029] The defect data sample set is used as the input of the decision tree model, and the printing operation parameter sample set is used as the output of the decision tree model to construct the printing parameter prediction model.

[0030] In one embodiment, before the step of optimizing the initial patch operation parameters and the initial printing operation parameters according to the intrinsic relationship between the patch operation parameters and the printing operation parameters by the parameter optimization model trained by the preset multi-task learning framework model to obtain the target patch operation parameters and the target printing operation parameters, the method further includes:

[0031] Obtaining a sample set of placement operation parameters and a sample set of printing operation parameters of a printed circuit board assembly production system;

[0032] The multi-task learning framework model is iteratively trained according to the patch operation parameter sample set and the printing operation parameter sample set to obtain the parameter optimization model.

[0033] In addition, to achieve the above-mentioned purpose, the present application also provides a device for determining process parameters of a printed circuit board assembly, the device comprising:

[0034] An acquisition module, used for acquiring defect data of a target printed circuit board assembly;

[0035] A parameter determination module, used for inputting the defect data into a pre-trained patch parameter prediction model and a printing parameter prediction model, respectively, to obtain initial patch operation parameters and initial printing operation parameters;

[0036] A parameter optimization module, for optimizing the initial patch operation parameters and the initial printing operation parameters according to the intrinsic relationship between the patch operation parameters and the printing operation parameters through a preset parameter optimization model trained based on a multi-task learning framework model, so as to obtain target patch operation parameters and target printing operation parameters;

[0037] The parameter adjustment module is used to adjust the patch operation parameters and the printing operation parameters of the printed circuit board assembly production system to the target patch operation parameters and the target printing operation parameters.

[0038] In addition, to achieve the above-mentioned purpose, the present application also provides a device, which is an electronic device, and the electronic device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, and the computer program is configured to implement the steps of the method for determining the process parameters of the printed circuit board assembly as described above.

[0039] In addition, to achieve the above-mentioned purpose, the present application also provides a medium, which is a computer-readable storage medium, and a computer program is stored on the computer-readable storage medium. When the computer program is executed by a processor, the steps of the method for determining the process parameters of the printed circuit board assembly as described above are implemented.

[0040] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, including a computer program, which implements the steps of the method for determining process parameters of a printed circuit board assembly as described above when executed by a processor.

[0041] The present application provides a method for determining process parameters of a printed circuit board assembly, firstly obtaining defect data of a target printed circuit board assembly; then inputting the defect data into a pre-trained patch parameter prediction model and a printing parameter prediction model respectively, so as to use the patch parameter prediction model and the printing parameter prediction model to preliminarily determine the initial patch operation parameters and initial printing operation parameters that the printed circuit board assembly production system should adjust to in order to overcome the defects of the produced printed circuit board assembly; then, by using a preset parameter optimization model trained based on a multi-task learning framework model, according to the inherent relationship between the patch operation parameters and the printing operation parameters, the initial patch operation parameters and the initial printing operation parameters are optimized to obtain the target patch operation parameters and the target printing operation parameters. Since there is a certain relationship between the patch operation parameters and the printing operation parameters of the printed circuit board assembly production system, they will affect each other, and since the multi-task learning framework model can learn multiple related tasks at the same time, the relationship between the tasks can be analyzed. Therefore, by using the parameter optimization model trained based on the multi-task learning framework model, that is, by using the parameter optimization model that integrates the intrinsic connection between the patch operation parameters and the printing operation parameters, the initial patch operation parameters and the initial printing operation parameters can be optimized, thereby obtaining the target patch operation parameters and the target printing operation parameters that take into account the synergy between the operation parameters. Afterwards, by adjusting the patch operation parameters and the printing operation parameters of the printed circuit board assembly production system to the target patch operation parameters and the target printing operation parameters, the production defects existing in the printed circuit board assembly production system can be better overcome to achieve the production optimization of the printed circuit board assembly.

[0042] Therefore, the present application provides a method for automatically optimizing the production of printed circuit board assemblies, wherein the entire process does not require manual intervention, and the correlation between patch operation parameters and printing operation parameters is fully considered during the optimization process. Therefore, compared with conventional manual optimization methods, not only the production optimization efficiency of printed circuit board assembly products is improved, but also the production optimization effect of printed circuit board assembly products is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0044] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0045] Figure 1A schematic flow chart of a method for determining process parameters of a printed circuit board assembly provided in the first embodiment of the present application;

[0046] Figure 2 A schematic diagram of the implementation principle of the parameter optimization model provided in the embodiment of the present application;

[0047] Figure 3 An architectural diagram of the convolutional layer of the model to be trained provided in an embodiment of the present application;

[0048] Figure 4 A schematic diagram of the principle of feature extraction using variable kernel convolution provided in an embodiment of the present application;

[0049] Figure 5 A schematic diagram of the distribution of components in a region of a printed circuit board assembly provided in an embodiment of the present application;

[0050] Figure 6 A schematic diagram of the overall implementation process of the method for determining process parameters of a printed circuit board assembly provided in an embodiment of the present application;

[0051] Figure 7 A schematic diagram of the module structure of a device for determining process parameters of a printed circuit board assembly provided in an embodiment of the present application;

[0052] Figure 8 A schematic diagram of the structure of the hardware operating environment involved in the embodiments of the present application.

[0053] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0054] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0055] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0056] As one of the core links in the assembly of electronic products, the SMT workshop is usually equipped with printing machines, placement machines, reflow soldering, AOI and other equipment. These equipment are mainly responsible for mounting electronic components (such as chips, resistors, capacitors, etc.) on printed circuit boards to form printed circuit board assemblies.

[0057] However, each device in the SMT workshop operates independently, and the parameters of each device fluctuate greatly during the production of printed circuit board assembly products. On this basis, in order to ensure the consistency and quality of the produced printed circuit board assembly products, manual debugging is usually performed based on the operator's experience, which is not only inefficient for the production optimization of printed circuit board assembly products, but also difficult to achieve a good production optimization effect.

[0058] Based on this, the present application provides a method for determining the process parameters of a printed circuit board assembly, first obtaining the defect data of the target printed circuit board assembly; then inputting the defect data into the pre-trained patch parameter prediction model and the printing parameter prediction model respectively, so as to use the patch parameter prediction model and the printing parameter prediction model to preliminarily determine the initial patch operation parameters and the initial printing operation parameters that the printed circuit board assembly production system should adjust to in order to overcome the defects of the produced printed circuit board assembly; then, through the preset parameter optimization model trained based on the multi-task learning framework model, according to the inherent relationship between the patch operation parameters and the printing operation parameters, the initial patch operation parameters and the initial printing operation parameters are optimized to obtain the target patch operation parameters and the target printing operation parameters. Since there is a certain relationship between the patch operation parameters and the printing operation parameters of the printed circuit board assembly production system, they will affect each other, and since the multi-task learning framework model can learn multiple related tasks at the same time, it is possible to analyze the relationship between the tasks. Therefore, by using the parameter optimization model trained based on the multi-task learning framework model, that is, by using the parameter optimization model that integrates the intrinsic connection between the patch operation parameters and the printing operation parameters, the initial patch operation parameters and the initial printing operation parameters can be optimized, thereby obtaining the target patch operation parameters and the target printing operation parameters that take into account the synergy between the operation parameters. Afterwards, by adjusting the patch operation parameters and the printing operation parameters of the printed circuit board assembly production system to the target patch operation parameters and the target printing operation parameters, the production defects existing in the printed circuit board assembly production system can be better overcome to achieve the production optimization of the printed circuit board assembly.

[0059] Therefore, the present application provides a method for automatically optimizing the production of printed circuit board assemblies, wherein the entire process does not require manual intervention, and the correlation between patch operation parameters and printing operation parameters is fully considered during the optimization process. Therefore, compared with conventional manual optimization methods, not only the production optimization efficiency of printed circuit board assembly products is improved, but also the production optimization effect of printed circuit board assembly products is improved.

[0060] The executor of the method for determining the process parameters of the printed circuit board assembly of the present application can be an electronic device with data processing, network communication and program running functions, such as a mobile phone, a tablet computer, a computer, etc.; it can also be a control device, a control system, a control circuit, etc. that can realize data processing, network communication and program running functions; it can also be a printed circuit board assembly production system, which is not specifically limited in this embodiment.

[0061] The following embodiments are described below by taking an electronic device as an example.

[0062] Based on this, the present application proposes a method for determining process parameters of a printed circuit board assembly according to a first embodiment. Figure 1 The method for determining the process parameters of a printed circuit board assembly includes steps S10 to S40:

[0063] Step S10, acquiring defect data of a target printed circuit board assembly;

[0064] It should be noted that the target PCB assembly is used as a reference PCB assembly when optimizing the production parameters (i.e., patch operation parameters and printing operation parameters) of the PCB assembly production system. The target PCB assembly can be a PCB assembly currently being produced by the PCB assembly production system. The PCB assembly production system fully covers the integrated and efficient production process performed in the STM workshop. The defect data may include but is not limited to the defect type (such as excessive tin, insufficient tin, and continuous tin) of the target PCB assembly and the regional location of the defect.

[0065] Step S20, inputting the defect data into the pre-trained patch parameter prediction model and the printing parameter prediction model respectively to obtain initial patch operation parameters and initial printing operation parameters;

[0066] It should be noted that the patch parameter prediction model is used to predict what patch operation parameters can be used to overcome the production defects existing in the printed circuit board assembly production system. The printing parameter prediction model is used to predict what printing operation parameters can be used to overcome the production defects existing in the printed circuit board assembly production system. The patch operation parameters refer to the operating parameters of the patch machine in the printed circuit board assembly production system, such as the patch speed, nozzle suction, placement position, etc. of the patch machine; the printing operation parameters refer to the operating parameters of the printer in the printed circuit board assembly production system, such as the printing speed, scraper pressure, demolding speed, etc. of the printer. The initial patch operation parameters are: when the inherent connection between the patch operation parameters and the printing operation parameters is not considered, in order to overcome the production defects existing in the printed circuit board assembly production system, the patch operation parameters that the printed circuit board assembly production system needs to adjust to. The initial printing operation parameters are: when the inherent connection between the patch operation parameters and the printing operation parameters is not considered, in order to overcome the production defects existing in the printed circuit board assembly production system, the printing operation parameters that the printed circuit board assembly production system needs to adjust to.

[0067] In addition, it should be noted that the patch parameter prediction model and the printing parameter prediction model essentially determine the causes of production defects in the printed circuit board assembly production system by analyzing defect data, and then use the solutions corresponding to the analyzed causes as the output of the model to obtain the initial patch operation parameters and initial printing operation parameters. Among them, there is a certain correlation between defects and the patch operation parameters and printing operation parameters used to overcome the defects. For example, assuming that the patch parameter prediction model determines that the production defects of the printed circuit board assembly production system are caused by too fast patch speed by analyzing defect data, a smaller patch speed can be output as the initial patch operation parameter. The printing parameter prediction model determines that the production defects of the printed circuit board assembly production system are caused by too small scraper pressure by analyzing defect data, and a larger scraper pressure can be output as the initial printing operation parameter.

[0068] In a feasible implementation, the training process of the patch parameter prediction model and the printing parameter prediction model may include steps S21 to S23:

[0069] Step S21, obtaining a defect data sample set of a printed circuit board assembly, a patch operation parameter sample set and a printing operation parameter sample set of a printed circuit board assembly production system;

[0070] It should be noted that the defect data sample set consists of multiple defect data samples, and one defect data sample records the relevant data of a defect image sample of a printed circuit board assembly; the defect data sample is the defect data used to train the relevant model. The patch operation parameter sample set consists of multiple patch operation parameter samples, and the patch operation parameter samples are the patch operation parameters used to train the relevant model. The printing operation parameter sample set consists of multiple printing operation parameter samples, and the printing operation parameter samples are the printing operation parameters used to train the relevant model.

[0071] In order to ensure the prediction accuracy of the subsequently constructed patch parameter prediction model and printing parameter prediction model, the acquired defect data sample set, patch operation parameter sample set and printing operation parameter sample set can be first subjected to data cleaning processing to filter out duplicate data and abnormal data in the three sample sets; and then the defect data sample set, patch operation parameter sample set and printing operation parameter sample set after data cleaning processing can be used to construct the patch parameter prediction model and the printing parameter prediction model.

[0072] Step S22, using the defect data sample set as the input of a preset decision tree model, and using the patch operation parameter sample set as the output of the decision tree model, so as to construct a patch parameter prediction model;

[0073] It should be noted that the decision tree model is a kind of decision making for data classification or regression by building a tree structure. Taking the defect data sample set as the input of the decision tree model and the patch operation parameter sample set as the output of the decision tree model actually means taking the defect data sample set as the feature matrix x and the patch operation parameter sample set as the target variable y1, and inputting them into the decision tree model for training to build a patch parameter prediction model.

[0074] Step S23, using the defect data sample set as the input of the decision tree model, and using the printing operation parameter sample set as the output of the decision tree model, so as to construct a printing parameter prediction model.

[0075] It should be noted that using the defect data sample set as the input of the decision tree model and the printing operation parameter sample set as the output of the decision tree model actually means using the defect data sample set as the feature matrix x and the printing operation parameter sample set as the target variable y2, which are input into the decision tree model for training to construct a printing parameter prediction model.

[0076] In other feasible implementations, the decision tree model in step S22 and step S23 may be replaced with a machine learning model such as a random forest, a support vector machine, or a neural network, etc., so as to use other models as a basic architecture or starting point to construct a patch parameter prediction model and a printing parameter prediction model. This embodiment does not specifically limit the specific construction method of the patch parameter prediction model and the printing parameter prediction model.

[0077] Step S30, optimizing the initial patch operation parameters and the initial printing operation parameters according to the intrinsic relationship between the patch operation parameters and the printing operation parameters through the preset parameter optimization model trained based on the multi-task learning framework model, and obtaining the target patch operation parameters and the target printing operation parameters;

[0078] It should be noted that the intrinsic connection between the patch operation parameters and the printing operation parameters includes the continuity of the two operations of patch and printing in the process flow (for example, if the printing operation parameters such as printing pressure and speed change, it may cause uneven coating, inconsistent thickness or position shift of the solder paste, thereby affecting the operation effect of the patch machine), the mutual dependence between the two parameters (for example, if the printing speed changes, it may cause different drying speeds of the solder paste, thereby affecting the adsorption and positioning effect of the patch machine on the components during the patch process; for example, if the printing pressure changes, it may affect the viscosity and fluidity of the solder paste, thereby indirectly affecting the patch effect of the patch machine), etc. The target patch operation parameters are: when considering the intrinsic connection between the patch operation parameters and the printing operation parameters, in order to overcome the production defects existing in the printed circuit board assembly production system, the patch operation parameters that the printed circuit board assembly production system needs to adjust to. The target printing operation parameters are: when considering the intrinsic connection between the patch operation parameters and the printing operation parameters, in order to overcome the production defects existing in the printed circuit board assembly production system, the printing operation parameters that the printed circuit board assembly production system needs to adjust to.

[0079] In addition, it should be noted that the multi-task learning framework (Progressive Layered Extraction, PLE) model can learn multiple related tasks at the same time to analyze the relationship between the tasks. Therefore, the parameter optimization model trained based on the multi-task learning framework model can integrate and analyze the internal relationship between the patch operation parameters and the printing operation parameters, so as to optimize the initial patch operation parameters and the initial printing operation parameters.

[0080] Among them, the parameter optimization model trained based on the multi-task learning framework model is used to optimize the input initial patch operation parameters and initial printing operation parameters. The specific implementation principle of obtaining the target patch operation parameters and target printing operation parameters can be referred to Figure 2 .

[0081] In a feasible implementation, the specific process of the parameter optimization model obtained by training the multi-task learning framework model may include steps S31 to S32:

[0082] Step S31, obtaining a sample set of patch operation parameters and a sample set of printing operation parameters of a printed circuit board assembly production system;

[0083] Step S32, iteratively training the multi-task learning framework model according to the patch operation parameter sample set and the printing operation parameter sample set to obtain a parameter optimization model.

[0084] Among them, in the process of iterative training of the multi-task learning framework model, the model can be further optimized through cross-validation and hyperparameter adjustment to ensure that the parameter optimization model finally trained can more accurately analyze the intrinsic relationship between the patch operation parameters and the printing operation parameters, thereby better optimizing the patch operation parameters and the printing operation parameters.

[0085] Step S40, adjusting the placement operation parameters and printing operation parameters of the printed circuit board assembly production system to target placement operation parameters and target printing operation parameters.

[0086] It is understandable that defects in printed circuit board assemblies are usually caused by inappropriate patch operation parameters and printing operation parameters used in the printed circuit board assembly production system. Therefore, by adjusting the patch operation parameters and printing operation parameters used in the printed circuit board assembly production system to appropriate patch operation parameters and printing operation parameters, that is, target patch operation parameters and target printing operation parameters, the production defects in the printed circuit board assembly production system can be overcome.

[0087] The present embodiment provides a method for determining the process parameters of a printed circuit board assembly, firstly obtaining the defect data of the target printed circuit board assembly; then inputting the defect data into the pre-trained patch parameter prediction model and the printing parameter prediction model respectively, so as to use the patch parameter prediction model and the printing parameter prediction model to preliminarily determine the initial patch operation parameters and the initial printing operation parameters that the printed circuit board assembly production system should adjust to in order to overcome the defects of the produced printed circuit board assembly; then, by using the preset parameter optimization model trained based on the multi-task learning framework model, according to the inherent relationship between the patch operation parameters and the printing operation parameters, the initial patch operation parameters and the initial printing operation parameters are optimized to obtain the target patch operation parameters and the target printing operation parameters. Since there is a certain relationship between the patch operation parameters and the printing operation parameters of the printed circuit board assembly production system, they will affect each other, and since the multi-task learning framework model can learn multiple related tasks at the same time, the relationship between the tasks can be analyzed. Therefore, by using the parameter optimization model trained based on the multi-task learning framework model, that is, by using the parameter optimization model that integrates the intrinsic connection between the patch operation parameters and the printing operation parameters, the initial patch operation parameters and the initial printing operation parameters can be optimized, thereby obtaining the target patch operation parameters and the target printing operation parameters that take into account the synergy between the operation parameters. Afterwards, by adjusting the patch operation parameters and the printing operation parameters of the printed circuit board assembly production system to the target patch operation parameters and the target printing operation parameters, the production defects existing in the printed circuit board assembly production system can be better overcome to achieve the production optimization of the printed circuit board assembly.

[0088] Therefore, this embodiment provides a method for automatically optimizing the production of printed circuit board assemblies. The entire process does not require manual participation, and the correlation between the patch operation parameters and the printing operation parameters is fully considered during the optimization process. Therefore, compared with the conventional manual optimization method, it not only improves the production optimization efficiency of printed circuit board assembly products, but also improves the production optimization effect of printed circuit board assembly products.

[0089] With the development of electronic technology, the density of components in modern PCB design is gradually increasing. At the same time, in order to ensure that the overall PCB circuit can meet the requirements of function, performance, reliability, etc., engineers usually use components of different sizes based on different power consumption, frequency or electrical performance requirements during the PCB design process. The welding quality of these components of different sizes not only affects the performance and stability of the circuit, but also directly affects the service life and maintenance cost of the product. Therefore, defect detection of PCBA is not only an important part of the electronic manufacturing process, but also a key step to ensure product quality, improve production efficiency and reduce costs.

[0090] Common PCBA defect detection technologies include manual visual inspection and AOI inspection. Manual visual inspection means that the operator uses the naked eye to check the PCBA welding and component placement; AOI inspection uses high-resolution camera systems and image processing algorithms to automatically detect PCBA welding problems (offset, insufficient tin, short circuit, etc.).

[0091] By improving the AOI image processing algorithm, the efficiency of PCBA defect detection can be further improved. With the development of technology, artificial intelligence (AI) technology is increasingly used in PCBA defect detection, such as using the classic convolutional neural network architecture to extract features from PCBA images and classify and locate defects, using classic target detection models or instance segmentation models (such as YOLO, etc.) to detect and locate defects in PCBA images, etc. Although PCBA defect detection technology based on deep learning has developed rapidly, defect detection of components of different scales for PCBA still needs to be improved. When the camera is at the same height for shooting, the components in the same FOV (Field of View) frame may have problems such as unbalanced distribution of component types, large size differences, and unclear features, which affect the performance of detection.

[0092] Based on this and the above-mentioned first embodiment, a second embodiment of the method for determining process parameters of a printed circuit board assembly of the present application is proposed. In the second embodiment, step S10 may include steps S11 to S12:

[0093] Step S11, acquiring an image of a target printed circuit board assembly;

[0094] Step S12, inputting the image of the target printed circuit board assembly into a pre-trained defect detection model to obtain defect data of the target printed circuit board assembly;

[0095] Among them, the defect detection model is trained based on the model to be trained, the backbone network of the model to be trained introduces variable kernel convolution, and the positioning loss function of the model to be trained is the pixel intersection-union ratio loss function.

[0096] It should be noted that the defect detection model is used to detect defects in printed circuit board assemblies. The model to be trained provides an infrastructure or starting point for the defect detection model. The model to be trained can be a YOLOv8 model, which is the eighth generation target detection model in the YOLO (You Only Look Once) series. Arbitrary Kernel Convolution (AKConv) is a convolution kernel with a variable shape. By introducing the variable kernel convolution into the backbone network (Backbone) of the model to be trained, the model's ability to extract multi-scale component features can be effectively improved.

[0097] For example, assuming that the model to be trained is a YOLOv8 model, the fourth and fifth ordinary convolutions in the backbone network of the YOLOv8 model are replaced by variable kernel convolutions (i.e., AKConv), then the convolution layer of the model to be trained will be presented as Figure 3 The architecture shown in the figure (BConv is a common convolution). After the variable kernel convolution is introduced into the backbone network of the YOLOv8 model, the process of using variable kernel convolution for feature extraction can be referred to in detail. Figure 4 (The image to be extracted features is processed by a series of processes such as convolution function Conv2d and resampling function Resample, and then the features in the image can be extracted.) Figure 4 It can be seen that compared with ordinary convolution with a fixed convolution kernel size, variable kernel convolution can select the appropriate convolution kernel size and shape according to the characteristics of the input image data in each layer of convolution operation. Therefore, by dynamically selecting the convolution kernel, the model can better adapt to the features of different scales, shapes and high feature complexity in the component defect detection task, thereby improving the recognition and generalization capabilities of the model. In addition, unlike the ordinary convolution, where the number of parameters grows quadratically, the number of parameters of the variable kernel convolution changes linearly, so the complexity of the model can be further reduced by dynamically adjusting the number of parameters.

[0098] In addition, it should be noted that the Pixels-IoU (PIoU) loss function can help the model solve the problem of imbalanced data sets. Specifically, the Pixels-IoU loss function can obtain a more refined regression bounding box through pixel-level overlap, and it can provide a more stable gradient to help the model improve the training speed and reduce the possibility of falling into the local optimum, thereby further improving the detection accuracy of the model. It is understandable that the components on the printed circuit board assembly are often of different sizes and densely distributed (see Figure 5 , the sizes of components A1~A5 in the figure are the same), introducing the pixel intersection-over-union loss function can help the model more accurately locate some small, tilted or specially shaped components, thereby improving the accuracy and comprehensiveness of defect detection.

[0099] Among them, the calculation formula of the pixel intersection-over-union loss function is as follows:

[0100]

[0101] Among them, b represents the real sample, b' represents the predicted sample border, and |M| represents the number of all positive samples.

[0102] This embodiment introduces variable kernel convolution into the backbone network of the model to be trained, and adjusts the positioning loss function of the model to be trained to the pixel intersection-over-union loss function, so as to improve the multi-scale component feature extraction capability and defect detection accuracy of the defect detection model trained by the improved model to be trained through the improved model to be trained, thereby solving the problems of missed detection and low detection accuracy in conventional PCBA defect detection methods due to large differences in component sizes and unclear features. It not only improves the accuracy of defect detection, but also ensures the comprehensiveness of defect detection, thereby further improving the production optimization effect of subsequent printed circuit board assembly products.

[0103] In a feasible implementation manner, the specific process of obtaining a defect detection model based on the training model may include steps S01 to S02:

[0104] Step S01, obtaining a defect data sample set of a printed circuit board assembly;

[0105] In a feasible implementation, step S01 may include steps S011 to S013:

[0106] Step S011, obtaining a plurality of defect image samples of a printed circuit board assembly;

[0107] It should be noted that the defect image sample is a defect image used to prepare a defect data sample set. The defect image refers to an image of a defective printed circuit board assembly. Generally speaking, the multiple defect image samples obtained need to include at least six defect types: offset, sideways, tombstone, flip, insufficient tin, and short circuit.

[0108] Step S012, marking the defect area and defect type of each defect image sample to construct an initial defect data set of the printed circuit board assembly;

[0109] It should be noted that the LableImg software can be used to annotate the defect area and defect type of each defect image sample, and generate an xml file in VOC (Visual Object Classes) format or other txt files that support model training, so as to construct an initial defect data set for the printed circuit board assembly.

[0110] Step S013: determining a defect data sample set according to the initial defect data set.

[0111] It should be noted that the initial defect data set is used as a benchmark for determining the defect data sample set.

[0112] In a feasible implementation, step S013 may include steps S014 to S015:

[0113] Step S014, performing size unification processing on each image included in the initial defect data set;

[0114] It should be noted that the images included in the initial defect data set refer to defect image samples after the defect area and defect type are labeled. The images included in the initial defect data set can be cropped into the same size by an image cropping algorithm to achieve size unification of the images included in the initial defect data set.

[0115] Step S015, performing data enhancement processing on the initial defect data set after the size unification processing to obtain a defect data sample set.

[0116] It should be noted that the data enhancement processing performed on the initial defect data set after the size unification processing may include but is not limited to flipping, rotating, cropping, adjusting brightness, contrast, etc.

[0117] In this implementation, first, the images included in the initial defect data set are resized to ensure that the subsequent model to be trained receives inputs of consistent size during the training process, thereby avoiding the additional complexity introduced by different image sizes, thereby improving the robustness and generalization ability of the defect detection model finally trained. Then, the initial defect data set after resizing is subjected to data enhancement processing to increase the diversity of defect data in the final defect data sample set, thereby further improving the robustness and generalization ability of the defect detection model finally trained.

[0118] In other feasible implementations of step S013, the initial defect data set may be directly used as the defect data sample set, which is not specifically limited in this embodiment.

[0119] Step S02: iteratively train the model to be trained based on the defect data sample set to obtain a defect detection model.

[0120] It should be noted that, in the process of iteratively training the model to be trained based on the defect data sample set to obtain the defect detection model, the defect data sample set can be first divided into a training set, a validation set and a test set according to a certain split ratio (for example, 8:1:1); then the training set is used to train the model to be trained, and after each training, the validation set is used to further adjust the model's hyperparameters to prevent the model from overfitting during the training process, and the test set is also used to evaluate the performance and generalization ability of the constructed model until the trained model is detected to converge, and the defect detection model can be obtained.

[0121] In a feasible implementation, step S02 may include steps S021 to S023:

[0122] Step S021, freezing a first preset number of convolutional layers close to the input layer in the model to be trained;

[0123] It should be noted that the first preset number refers to the number of convolutional layers that need to be frozen. The first preset number can be a default value or can be flexibly set by the user according to actual conditions. This embodiment does not specifically limit this.

[0124] Step S022, performing iterative training for a second preset number of training cycles on the frozen model to be trained according to the defect data sample set to obtain an initial detection model;

[0125] It should be noted that the second preset number refers to the number of training cycles (epochs) required for the frozen model to be trained. The second preset number can be a default value, such as 50; it can also be flexibly set by the user according to actual conditions, and this embodiment does not make specific limitations on this.

[0126] Step S023, unfreeze the initial detection model, and perform iterative training for a third preset number of training cycles on the unfrozen initial detection model based on the defect data sample set to obtain a defect detection model.

[0127] It should be noted that the third preset number refers to the number of training cycles required for the initial detection model after thawing. The third preset number can be a default value, such as 100; it can also be flexibly set by the user according to actual conditions, and this embodiment does not make specific limitations on this.

[0128] It can be understood that the features learned by several convolutional layers close to the input layer (such as edges, textures, etc.) are often universal and have good applicability for different tasks. Therefore, this embodiment first freezes the first preset number of convolutional layers close to the input layer in the model to be trained, and then iteratively trains the frozen model to be trained for a second preset number of training cycles based on the defect data sample set to obtain an initial detection model; thereby, the stability of these universal features in the initial stage of model training can be ensured to avoid the problem of model overfitting when the defect data sample set has less data. Afterwards, the initial detection model is unfrozen, and the unfrozen initial detection model is iteratively trained for a third preset number of training cycles based on the defect data sample set to obtain a defect detection model. In this way, the weights of the previously frozen convolutional layers can be updated to make the model more suitable for the target task, so as to improve the detection accuracy of the defect detection model finally trained.

[0129] In a feasible implementation manner, step S02 may further include steps S024 to S025:

[0130] Step S024, performing data enhancement processing on the defect data sample set to obtain a defect data sample set after data enhancement processing;

[0131] It should be noted that data enhancement processing for the defect data sample set may include but is not limited to Mosaic data enhancement, Mixup data enhancement, etc. Among them, Mosaic is an image data enhancement technology that increases the diversity of data by splicing four images into one image. It not only increases the number of training samples of the model, but also introduces more contextual information through the combination of different images, which helps to improve the generalization ability of the model. Mixup is also an image data enhancement technology that generates new training samples by performing linear interpolation between two randomly selected samples. Specifically, it randomly selects two samples (images and their labels), and then mixes the pixel values ​​and labels of the two samples in a certain proportion. This not only increases the diversity of training data, but also prompts the model to learn smoother decision boundaries during training, thereby improving the generalization ability of the model.

[0132] Step S025, iteratively training the model to be trained based on the defect data sample set after data enhancement processing to obtain a defect detection model.

[0133] In this embodiment, by performing data enhancement processing on the defect data sample set, the defect data sample set after data enhancement processing is used to iteratively train the training model to obtain the defect detection model, thereby improving the generalization ability of the defect detection model finally obtained by training.

[0134] For example, to help understand the implementation process of the method for determining the process parameters of a printed circuit board assembly obtained by combining the above embodiments, take the training model as the YOLOv8 model as an example, please refer to Figure 6 , specifically:

[0135] Firstly, a defect data sample set of printed circuit board assemblies is obtained; then, the ordinary convolution of the YOLOv8 model is replaced by a variable kernel convolution, and the positioning loss function of the YOLOv8 model is replaced by a pixel intersection-over-union loss function to improve the YOLOv8 model; then, based on the defect data sample set, the improved YOLOv8 model is iteratively trained to obtain a defect detection model, and the defect detection model is deployed on the production line to detect the defect data of the target printed circuit board assembly using the defect detection model; then, a patch parameter prediction model and a printing parameter prediction model are constructed based on the defect data sample set, the patch operation parameter sample set and the printing operation parameter sample set; then, the defect detection model is used to detect the defect data of the target printed circuit board assembly. The defect data of the standard printed circuit board assembly are respectively input into the patch parameter prediction model and the printing parameter prediction model to obtain the initial patch operation parameters and the initial printing operation parameters; then, the initial patch operation parameters and the initial printing operation parameters are input into the parameter optimization model trained based on the multi-task learning framework model, so as to optimize the initial patch operation parameters and the initial printing operation parameters according to the intrinsic relationship between the patch operation parameters and the printing operation parameters through the parameter optimization model, and obtain the target patch operation parameters and the target printing operation parameters; finally, the patch operation parameters and the printing operation parameters of the printed circuit board assembly production system are adjusted to the target patch operation parameters and the target printing operation parameters, so as to realize the production optimization of the printed circuit board assembly.

[0136] It should be noted that this example is only used to assist in understanding the present application and does not constitute a limitation on the method for determining process parameters of the printed circuit board assembly of the present application. More simple transformations based on this technical concept are all within the scope of protection of the present application.

[0137] The present application also provides a device for determining process parameters of a printed circuit board assembly. Figure 7 , a device for determining process parameters of a printed circuit board assembly comprises:

[0138] An acquisition module 10, used to acquire defect data of a target printed circuit board assembly;

[0139] The parameter determination module 20 is used to input the defect data into the pre-trained patch parameter prediction model and the printing parameter prediction model to obtain the initial patch operation parameters and the initial printing operation parameters;

[0140] The parameter optimization module 30 is used to optimize the initial patch operation parameters and the initial printing operation parameters according to the intrinsic relationship between the patch operation parameters and the printing operation parameters through the parameter optimization model trained by the preset multi-task learning framework model, so as to obtain the target patch operation parameters and the target printing operation parameters;

[0141] The parameter adjustment module 40 is used to adjust the patch operation parameters and the printing operation parameters of the printed circuit board assembly production system to target patch operation parameters and target printing operation parameters.

[0142] Optionally, the acquisition module 10 is further used for:

[0143] acquiring an image of a target printed circuit board assembly;

[0144] Inputting the image of the target printed circuit board assembly into a pre-trained defect detection model to obtain defect data of the target printed circuit board assembly;

[0145] Among them, the defect detection model is trained based on the model to be trained, the backbone network of the model to be trained introduces variable kernel convolution, and the positioning loss function of the model to be trained is the pixel intersection-union ratio loss function.

[0146] Optionally, the device for determining process parameters of a printed circuit board assembly further comprises:

[0147] Obtaining a sample set of defect data of a printed circuit board assembly;

[0148] According to the defect data sample set, the training model is iteratively trained to obtain a defect detection model.

[0149] Optionally, the device for determining process parameters of a printed circuit board assembly further comprises:

[0150] Freeze the first preset number of convolutional layers close to the input layer in the model to be trained;

[0151] According to the defect data sample set, the frozen model to be trained is iteratively trained for a second preset number of training cycles to obtain an initial detection model;

[0152] The initial detection model is unfrozen, and based on the defect data sample set, the unfrozen initial detection model is iteratively trained for a third preset number of training cycles to obtain a defect detection model.

[0153] Optionally, the device for determining process parameters of a printed circuit board assembly further comprises:

[0154] Performing data enhancement processing on the defect data sample set to obtain a defect data sample set after data enhancement processing;

[0155] According to the defect data sample set after data enhancement processing, the training model is iteratively trained to obtain the defect detection model.

[0156] Optionally, the device for determining process parameters of a printed circuit board assembly further comprises:

[0157] acquiring a plurality of defect image samples of a printed circuit board assembly;

[0158] Annotating the defect area and defect type of each defect image sample to construct an initial defect dataset of the printed circuit board assembly;

[0159] According to the initial defect data set, a defect data sample set is determined.

[0160] Optionally, the device for determining process parameters of a printed circuit board assembly further comprises:

[0161] Performing size unification processing on each image included in the initial defect data set;

[0162] The initial defect data set after size normalization is subjected to data enhancement processing to obtain a defect data sample set.

[0163] Optionally, the device for determining process parameters of a printed circuit board assembly further comprises:

[0164] Obtaining a sample set of defect data of a printed circuit board assembly, a sample set of mounting operation parameters of a printed circuit board assembly production system, and a sample set of printing operation parameters;

[0165] The defect data sample set is used as the input of the preset decision tree model, and the patch operation parameter sample set is used as the output of the decision tree model to build a patch parameter prediction model;

[0166] The defect data sample set is used as the input of the decision tree model, and the printing operation parameter sample set is used as the output of the decision tree model to build a printing parameter prediction model.

[0167] Optionally, the device for determining process parameters of a printed circuit board assembly further comprises:

[0168] Obtaining a sample set of placement operation parameters and a sample set of printing operation parameters of a printed circuit board assembly production system;

[0169] According to the patch operation parameter sample set and the printing operation parameter sample set, the multi-task learning framework model is iteratively trained to obtain a parameter optimization model.

[0170] The device for determining the process parameters of a printed circuit board assembly provided in the embodiment of the present application adopts the method for determining the process parameters of a printed circuit board assembly in the above embodiment, which can improve the production optimization efficiency and production optimization effect of the printed circuit board assembly product. Compared with the prior art, the beneficial effects of the device for determining the process parameters of a printed circuit board assembly provided in the embodiment of the present application are the same as the beneficial effects of the method for determining the process parameters of a printed circuit board assembly provided in the above embodiment, and are not described in detail here.

[0171] An embodiment of the present application also provides an electronic device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method for determining the process parameters of the printed circuit board assembly in the above embodiment.

[0172] Reference below Figure 8 , which shows a schematic diagram of the structure of an electronic device suitable for implementing an embodiment of the present application. Figure 8 The electronic device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0173] like Figure 8 As shown, the electronic device may include a processing device 101 (e.g., a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 102 or a program loaded from a storage device 103 to a random access memory (RAM: Random Access Memory) 104. In RAM104, various programs and data required for the operation of the electronic device are also stored. The processing device 101, ROM102, and RAM104 are connected to each other through a bus 105. An input / output (I / O) interface 106 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 106: an input device 107 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 108 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 103 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 109. The communication device 109 can allow the electronic device to communicate with other devices wirelessly or by wire to exchange data. Although the electronic device with various systems is shown in the figure, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or have instead.

[0174] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 103, or installed from a ROM 102. When the computer program is executed by the processing device 101, the above-mentioned functions defined in the method of the embodiment of the present application are executed.

[0175] The electronic device provided in the embodiment of the present application adopts the method for determining the process parameters of the printed circuit board assembly in the above embodiment, which can improve the production optimization efficiency and production optimization effect of the printed circuit board assembly product. Compared with the prior art, the beneficial effects of the electronic device provided in the embodiment of the present application are the same as the beneficial effects of the method for determining the process parameters of the printed circuit board assembly provided in the above embodiment, and other technical features in the electronic device are the same as the features disclosed in the above embodiment method, which will not be repeated here.

[0176] It should be understood that each part of the embodiments of the present application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0177] The above is only a specific implementation of the embodiment of the present application, but the protection scope of the embodiment of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or replacements within the technical scope disclosed in the present application, which should be included in the protection scope of the embodiment of the present application. Therefore, the protection scope of the embodiment of the present application shall be based on the protection scope of the above claims.

[0178] An embodiment of the present application further provides a computer-readable storage medium storing a computer program executable on a processor, wherein the computer program is used to execute the method for determining process parameters of a printed circuit board assembly in the above embodiment.

[0179] The computer-readable storage medium provided in the embodiment of the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM: Random Access Memory), a read-only memory (ROM: Read Only Memory), an erasable programmable read-only memory (EPROM: Erasable Programmable Read Only Memory or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM: CD-Read Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency: Radio Frequency), etc., or any suitable combination of the above.

[0180] The computer-readable storage medium may be included in the electronic device, or may exist independently without being installed in the electronic device.

[0181] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by an electronic device, the electronic device: obtains defect data of a target printed circuit board assembly; inputs the defect data into a pre-trained patch parameter prediction model and a printing parameter prediction model, respectively, to obtain initial patch operation parameters and initial printing operation parameters; optimizes the initial patch operation parameters and initial printing operation parameters according to the intrinsic relationship between the patch operation parameters and the printing operation parameters through a preset parameter optimization model trained based on a multi-task learning framework model, to obtain target patch operation parameters and target printing operation parameters; and adjusts the patch operation parameters and printing operation parameters of the printed circuit board assembly production system to the target patch operation parameters and target printing operation parameters.

[0182] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0183] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0184] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0185] The computer-readable storage medium provided in the embodiment of the present application stores computer-readable program instructions for executing the above-mentioned method for determining the process parameters of the printed circuit board assembly, which can improve the production optimization efficiency and production optimization effect of the printed circuit board assembly product. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in the embodiment of the present application are the same as the beneficial effects of the method for determining the process parameters of the printed circuit board assembly provided in the above-mentioned embodiment, and will not be repeated here.

[0186] An embodiment of the present application further provides a computer program product, including a computer program, which implements the steps of the method for determining process parameters of a printed circuit board assembly as described above when the computer program is executed by a processor.

[0187] The computer program product provided in the embodiment of the present application can improve the production optimization efficiency and production optimization effect of the printed circuit board assembly product. Compared with the prior art, the beneficial effects of the computer program product provided in the embodiment of the present application are the same as the beneficial effects of the method for determining the process parameters of the printed circuit board assembly provided in the above embodiment, and will not be repeated here.

[0188] The above are only preferred embodiments of the present application, and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent processing scope of the present application.

Claims

1. A method for determining process parameters of a printed circuit board assembly, characterized in that: The method comprises: acquiring defect data of a target printed circuit board assembly; Inputting the defect data into a pre-trained patch parameter prediction model and a printing parameter prediction model respectively to obtain initial patch operation parameters and initial printing operation parameters; By using a preset parameter optimization model trained based on a multi-task learning framework model, the initial patch operation parameters and the initial printing operation parameters are optimized according to the intrinsic relationship between the patch operation parameters and the printing operation parameters to obtain target patch operation parameters and target printing operation parameters; The mounting operation parameters and the printing operation parameters of the printed circuit board assembly production system are adjusted to the target mounting operation parameters and the target printing operation parameters.

2. The method for determining process parameters of a printed circuit board assembly according to claim 1, wherein: The step of obtaining defect data of the target printed circuit board assembly comprises: acquiring an image of the target printed circuit board assembly; Inputting the image of the target printed circuit board assembly into a pre-trained defect detection model to obtain defect data of the target printed circuit board assembly; Among them, the defect detection model is trained based on the model to be trained, the backbone network of the model to be trained introduces a variable kernel convolution, and the positioning loss function of the model to be trained is a pixel intersection-over-union loss function.

3. The method for determining process parameters of a printed circuit board assembly according to claim 2, wherein: The training process of the defect detection model includes: Obtaining a sample set of defect data of a printed circuit board assembly; Freezing a first preset number of convolutional layers close to an input layer in the model to be trained; According to the defect data sample set, performing iterative training for a second preset number of training cycles on the frozen model to be trained to obtain an initial detection model; The initial detection model is unfrozen, and based on the defect data sample set, the unfrozen initial detection model is iteratively trained for a third preset number of training cycles to obtain the defect detection model.

4. The method for determining process parameters of a printed circuit board assembly according to claim 3, wherein: The step of obtaining a defect data sample set of the printed circuit board assembly further includes: Performing data enhancement processing on the defect data sample set to obtain a defect data sample set after data enhancement processing.

5. The method for determining process parameters of a printed circuit board assembly according to claim 3, wherein: The step of obtaining a defect data sample set of a printed circuit board assembly comprises: acquiring a plurality of defect image samples of a printed circuit board assembly; Annotating the defect area and defect type of each defect image sample to construct an initial defect data set of the printed circuit board assembly; Performing size unification processing on each image included in the initial defect data set; The initial defect data set after size normalization is subjected to data enhancement processing to obtain the defect data sample set.

6. The method for determining process parameters of a printed circuit board assembly according to any one of claims 1 to 5, characterized in that: Before the step of inputting the defect data into a pre-trained patch parameter prediction model and a printing parameter prediction model to obtain initial patch operation parameters and initial printing operation parameters, the method further includes: Obtaining a sample set of defect data of a printed circuit board assembly, a sample set of mounting operation parameters of a printed circuit board assembly production system, and a sample set of printing operation parameters; Using the defect data sample set as input of a preset decision tree model, and using the patch operation parameter sample set as output of the decision tree model, so as to construct the patch parameter prediction model; The defect data sample set is used as the input of the decision tree model, and the printing operation parameter sample set is used as the output of the decision tree model to construct the printing parameter prediction model.

7. The method for determining process parameters of a printed circuit board assembly according to any one of claims 1 to 5, characterized in that: Before the step of optimizing the initial patch operation parameters and the initial printing operation parameters according to the intrinsic relationship between the patch operation parameters and the printing operation parameters by the parameter optimization model trained by the preset multi-task learning framework model to obtain the target patch operation parameters and the target printing operation parameters, the method further includes: Obtaining a sample set of placement operation parameters and a sample set of printing operation parameters of a printed circuit board assembly production system; The multi-task learning framework model is iteratively trained according to the patch operation parameter sample set and the printing operation parameter sample set to obtain the parameter optimization model.

8. A device for determining process parameters of a printed circuit board assembly, characterized in that: The device comprises: An acquisition module, used for acquiring defect data of a target printed circuit board assembly; A parameter determination module, used for inputting the defect data into a pre-trained patch parameter prediction model and a printing parameter prediction model, respectively, to obtain initial patch operation parameters and initial printing operation parameters; A parameter optimization module, for optimizing the initial patch operation parameters and the initial printing operation parameters according to the intrinsic relationship between the patch operation parameters and the printing operation parameters through a preset parameter optimization model trained based on a multi-task learning framework model, so as to obtain target patch operation parameters and target printing operation parameters; The parameter adjustment module is used to adjust the patch operation parameters and the printing operation parameters of the printed circuit board assembly production system to the target patch operation parameters and the target printing operation parameters.

9. A device, characterized in that: The device is an electronic device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the method for determining process parameters of a printed circuit board assembly as described in any one of claims 1 to 7.

10. A medium, characterized in that The medium is a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the method for determining process parameters of a printed circuit board assembly according to any one of claims 1 to 7 are implemented.

Images